Merge PR #135: Sync from CVXPY upstream master

Resolves merge conflicts:
- README.md: Keep DNLP-specific README
- cvxpy/atoms/pnorm.py: Keep DNLP _hess_vec methods, add upstream PnormApprox class
- cvxpy/utilities/citations.py: Keep IPOPT/UNO citations, add MOREAU citation

Co-Authored-By: Claude Opus 4.5 <[email protected]>

Author: Transurgeon

.github/workflows/docs.yml

@@ -36,15 +36,15 @@ jobs:
           echo DEPLOY_LATEST=$(python continuous_integration/doc_is_latest.py --c "$CURRENT_VERSION" --l "$LATEST_DEPLOYED_VERSION") >> $GITHUB_ENV
 
       - name: Deploy this version
-        uses: JamesIves/github-pages-deploy-action@v4.7.6
+        uses: JamesIves/github-pages-deploy-action@v4.8.0
         with:
           branch: gh-pages # The branch the action should deploy to.
           folder: doc/build/html # The folder the action should deploy.
           target-folder: ${{ env.VERSION_PATH }}
 
       - name: Deploy to root if latest version
         if: ${{env.DEPLOY_LATEST == 'True'}}
-        uses: JamesIves/github-pages-deploy-action@v4.7.6
+        uses: JamesIves/github-pages-deploy-action@v4.8.0
         with:
           branch: gh-pages # The branch the action should deploy to.
           folder: doc/build/html # The folder the action should deploy.

.github/workflows/scorecards.yml

@@ -51,6 +51,6 @@ jobs:
       # Upload the results to GitHub's code scanning dashboard (optional).
       # Commenting out will disable upload of results to your repo's Code Scanning dashboard
       - name: "Upload to code-scanning"
-        uses: github/codeql-action/upload-sarif@5d4e8d1aca955e8d8589aabd499c5cae939e33c7 # v2.16.4
+        uses: github/codeql-action/upload-sarif@19b2f06db2b6f5108140aeb04014ef02b648f789 # v2.16.4
         with:
           sarif_file: results.sarif

PROCEDURES.md

@@ -113,6 +113,13 @@ If this file has changed between versions, the old patch will fail to apply and
 ## Creating a release on GitHub
 Go to the [Releases](https://github.com/cvxpy/cvxpy/releases) tab and click "Draft a new release". Select the previously created tag and write release notes. For minor releases, this includes a summary of new features and deprecations. Additionally, we mention the PRs contained in the release and their contributors. Take care to select the "set as the latest release" only for minor releases or patches to the most recent major release.
 
+To generate the list of PRs and contributors, use the `tools/release_notes.py` script:
+```
+python tools/release_notes.py v1.8.0  # minor release
+python tools/release_notes.py v1.7.5  # patch release
+```
+For minor releases, the script automatically excludes PRs that were cherry-picked into the previous release branch's patch releases. For patch releases, it compares against the previous patch tag.
+
 ## Deploying updated documentation to gh-pages
 
 The web documentation is built and deployed using a GitHub action that can be found [here](https://github.com/cvxpy/cvxpy/blob/master/.github/workflows/docs.yml).

README.md

@@ -1,6 +1,6 @@
 # DNLP — Disciplined Nonlinear Programming
 The DNLP package is an extension of [CVXPY](https://www.cvxpy.org/) to general nonlinear programming (NLP).
-DNLP allows smooth functions to be freely mixed with nonsmooth convex and concave functions, 
+DNLP allows smooth functions to be freely mixed with nonsmooth convex and concave functions,
 with some rules governing how the nonsmooth functions can be used. For details, see our paper [Disciplined Nonlinear Programming](https://web.stanford.edu/~boyd/papers/dnlp.html).
 
 ---
@@ -27,7 +27,7 @@ pip install .
 Below we give a toy example where we maximize a convex quadratic function subject to a nonlinear equality constraint.  Many more examples, including the ones in the paper, can be found at [DNLP-examples](https://github.com/cvxgrp/dnlp-examples).
 ```python
 import cvxpy as cp
-import numpy as np 
+import numpy as np
 import cvxpy as cp
 
 # problem data

continuous_integration/install_dependencies.sh

@@ -17,7 +17,7 @@ fi
 
 uv pip install pytest pytest-cov hypothesis "setuptools>65.5.1"
 
-uv pip install scs clarabel osqp
+uv pip install scs clarabel osqp highspy
 
 if [[ "$RUNNER_OS" != "macOS" ]]; then
   uv pip install mkl

cvxpy/__init__.py

@@ -41,6 +41,9 @@
     enable_warnings as enable_warnings,
     warnings_enabled as warnings_enabled,
 )
+from cvxpy.utilities.warn import (
+    CvxpyDeprecationWarning as CvxpyDeprecationWarning,
+)
 from cvxpy.expressions.constants import (
     CallbackParam as CallbackParam,
     Constant as Constant,
@@ -59,6 +62,7 @@
     partial_optimize as partial_optimize,
     suppfunc as suppfunc,
 )
+from cvxpy import logic as logic
 from cvxpy.reductions.solvers.defines import installed_solvers as installed_solvers
 from cvxpy.settings import (
     CBC as CBC,

cvxpy/atoms/__init__.py

@@ -65,7 +65,7 @@
 from cvxpy.atoms.elementwise.minimum import minimum
 from cvxpy.atoms.elementwise.neg import neg
 from cvxpy.atoms.elementwise.pos import pos
-from cvxpy.atoms.elementwise.power import power
+from cvxpy.atoms.elementwise.power import Power, PowerApprox, power
 from cvxpy.atoms.elementwise.rel_entr import rel_entr
 from cvxpy.atoms.elementwise.scalene import scalene
 from cvxpy.atoms.elementwise.sqrt import sqrt
@@ -75,7 +75,7 @@
 from cvxpy.atoms.elementwise.hyperbolic import sinh, asinh, tanh, atanh
 from cvxpy.atoms.eye_minus_inv import eye_minus_inv, resolvent
 from cvxpy.atoms.gen_lambda_max import gen_lambda_max
-from cvxpy.atoms.geo_mean import geo_mean
+from cvxpy.atoms.geo_mean import GeoMean, GeoMeanApprox, geo_mean
 from cvxpy.atoms.gmatmul import gmatmul
 from cvxpy.atoms.harmonic_mean import harmonic_mean
 from cvxpy.atoms.inv_prod import inv_prod
@@ -98,7 +98,7 @@
 from cvxpy.atoms.one_minus_pos import diff_pos, one_minus_pos
 from cvxpy.atoms.perspective import perspective
 from cvxpy.atoms.pf_eigenvalue import pf_eigenvalue
-from cvxpy.atoms.pnorm import Pnorm, pnorm
+from cvxpy.atoms.pnorm import Pnorm, PnormApprox, pnorm
 from cvxpy.atoms.prod import Prod, prod
 from cvxpy.atoms.quad_form import QuadForm, quad_form
 from cvxpy.atoms.quad_over_lin import quad_over_lin
@@ -117,16 +117,18 @@
 # TODO(akshayka): Perhaps couple this information with the atom classes
 # themselves.
 SOC_ATOMS = [
-    geo_mean,
-    pnorm,
-    Pnorm,
+    GeoMeanApprox,
+    PnormApprox,
     QuadForm,
     quad_over_lin,
-    power,
+    PowerApprox,
     huber,
     std,
 ]
 
+POWCONE_ATOMS = [Pnorm, Power]
+POWCONE_ND_ATOMS = [GeoMean]
+
 EXP_ATOMS = [
     log_sum_exp,
     log_det,

cvxpy/atoms/affine/binary_operators.py

@@ -26,6 +26,7 @@
 import cvxpy.utilities as u
 from cvxpy.atoms.affine.add_expr import AddExpression
 from cvxpy.atoms.affine.affine_atom import AffAtom
+from cvxpy.atoms.affine.broadcast_to import broadcast_to
 from cvxpy.atoms.affine.conj import conj
 from cvxpy.atoms.affine.promote import Promote
 from cvxpy.atoms.affine.reshape import deep_flatten, reshape
@@ -131,6 +132,52 @@ class MulExpression(BinaryOperator):
     OP_NAME = "@"
     OP_FUNC = op.mul
 
+    def __init__(self, lh_exp, rh_exp) -> None:
+        # Broadcast batch dimensions for ND matmul
+        lh_exp, rh_exp = self._broadcast_batch_dims(lh_exp, rh_exp)
+        super(MulExpression, self).__init__(lh_exp, rh_exp)
+
+    @staticmethod
+    def _broadcast_batch_dims(lh_exp, rh_exp):
+        """
+        Broadcast batch dimensions for ND matrix multiplication.
+
+        For A @ B where A has shape (...a, m, k) and B has shape (...b, k, n),
+        broadcasts both to have batch shape broadcast(...a, ...b).
+        """
+        lh_exp = Expression.cast_to_const(lh_exp)
+        rh_exp = Expression.cast_to_const(rh_exp)
+
+        lh_shape = lh_exp.shape
+        rh_shape = rh_exp.shape
+
+        # Only apply batch broadcasting for ND arrays (ndim > 2)
+        if len(lh_shape) <= 2 and len(rh_shape) <= 2:
+            return lh_exp, rh_exp
+
+        # Extract batch dimensions (all but last 2)
+        lh_batch = lh_shape[:-2] if len(lh_shape) > 2 else ()
+        rh_batch = rh_shape[:-2] if len(rh_shape) > 2 else ()
+
+        # Compute broadcast batch shape
+        try:
+            broadcast_batch = np.broadcast_shapes(lh_batch, rh_batch)
+        except ValueError:
+            # Let shape validation handle the error with a clearer message
+            return lh_exp, rh_exp
+
+        # Broadcast lhs if needed
+        if lh_batch != broadcast_batch:
+            target_shape = broadcast_batch + lh_shape[-2:]
+            lh_exp = broadcast_to(lh_exp, target_shape)
+
+        # Broadcast rhs if needed
+        if rh_batch != broadcast_batch:
+            target_shape = broadcast_batch + rh_shape[-2:]
+            rh_exp = broadcast_to(rh_exp, target_shape)
+
+        return lh_exp, rh_exp
+
     def numeric(self, values):
         """Matrix multiplication.
         """
@@ -139,11 +186,6 @@ def numeric(self, values):
         else:
             return values[0] @ values[1]
 
-    def validate_arguments(self):
-        """Validate that the arguments can be multiplied together."""
-        if self.args[0].ndim > 2 or self.args[1].ndim > 2:
-            raise ValueError("Multiplication with N-d arrays is not yet supported")
-    
     def shape_from_args(self) -> Tuple[int, ...]:
         """Returns the (row, col) shape of the expression.
         """

cvxpy/atoms/affine/conv.py

@@ -14,7 +14,6 @@
 limitations under the License.
 """
 
-import warnings
 from typing import List, Tuple
 
 import numpy as np
@@ -26,6 +25,7 @@
 from cvxpy.atoms.affine.affine_atom import AffAtom
 from cvxpy.constraints.constraint import Constraint
 from cvxpy.expressions.constants.parameter import is_param_free
+from cvxpy.utilities.warn import CvxpyDeprecationWarning, warn
 
 
 class conv(AffAtom):
@@ -48,7 +48,7 @@ class conv(AffAtom):
     """
 
     def __init__(self, lh_expr, rh_expr) -> None:
-        warnings.warn("conv is deprecated. Use convolve instead.", DeprecationWarning)
+        warn("conv is deprecated. Use convolve instead.", CvxpyDeprecationWarning)
         super(conv, self).__init__(lh_expr, rh_expr)
 
     @AffAtom.numpy_numeric

cvxpy/atoms/affine/cumsum.py

@@ -13,7 +13,6 @@
 See the License for the specific language governing permissions and
 limitations under the License.
 """
-import warnings
 from typing import Optional, Tuple
 
 import numpy as np
@@ -23,6 +22,7 @@
 from cvxpy.atoms.affine.affine_atom import AffAtom
 from cvxpy.atoms.axis_atom import AxisAtom
 from cvxpy.expressions.expression import Expression
+from cvxpy.utilities.warn import warn
 
 
 def _sparse_triu_ones(dim: int) -> sp.csc_array:
@@ -58,7 +58,7 @@ def validate_arguments(self) -> None:
         """Validate axis, but handle 0D arrays specially."""
         if self.args[0].ndim == 0:
             if self.axis is not None:
-                warnings.warn(
+                warn(
                     "cumsum on 0-dimensional arrays currently returns a scalar, "
                     "but in a future CVXPY version it will return a 1-element "
                     "array to match numpy.cumsum behavior. Additionally, only "